035-Multimodal generative AI for medical image

This review explores multimodal generative AI (GenMI) for medical image interpretation, highlighting its potential to automate the creation of medical reports from images, especially within radiology. It discusses how GenMI, leveraging large language models (LLMs), could significantly reduce clinician workload, improve turnaround times, and enhance patient care and medical education by providing real-time, interactive expertise. However, the article also addresses formidable challenges, including the crucial need for rigorous validation of model accuracy, ensuring transparency, and mitigating biases inherent in current datasets and models, underscoring the importance of human oversight and continuous feedback in the deployment of such advanced AI systems.

References：

• Rao V M, Hla M, Moor M, et al. Multimodal generative AI for medical image interpretation[J]. Nature, 2025, 639(8056): 888-896.